As regards the detection of partial discharges, several methods have been developed, based on the use of different physical phenomena associated with discharges, such as, for example, methods of optical, acoustic and electrical type.
This invention relates in particular to detection methods of the electrical type which, as is known, involve measuring the current pulses that travel a detection circuit coupled to the electrical system being checked.
These detected current pulses (hereinafter referred to, for convenience, as discharge pulses) have a time profile that depends on the dynamics of the partial discharges that is, on the physics of the discharge phenomena) and on the nature of the means which the detected pulses cross as they travel from the discharge site (where the discharges occur) to the detection site. The time profile of the discharge pulses, consisting of the waveform of the pulses themselves, contains precious diagnostic information regarding both the physical phenomena associated with the discharges (correlated with the nature of the defects of the insulating system) and the nature of the medium which the detected pulses travel through (correlated with the location of the defects in the insulating system).
As to the difficulties of interpreting the results of partial discharge measurements, these depend on the fact that the measured data may be unreliable or insignificant.
In effect, during detection of the signals associated with the partial discharges, information essential for subsequent processing of the signals themselves for diagnostic purposes may be lost (loss of information might consist, for example, of failure to detect a pulse or failure to detect the waveform of a pulse).
Considering that the discharge pulses contain information that can be used at diagnostic level, it would be very important to detect the pulses in a very wide detection band, for example in the order of ten MHz or more, or at least several MHz.
It should be noted, however, that the signals detected are current pulses which are correlated with load transfers that occur in the defects in the electrical apparatus being checked and constitute the partial discharges, but they do not constitute a direct measurement of the load transfers.
Thus, to be able to estimate the intensity of a partial discharge corresponding to a signal detected, it is necessary to calibrate the acquisition instrument in order to determine the amplitude of the signals measured by the instrument and the load transfer associated with the partial discharges corresponding to those signals.
For that purpose, a calibrator is applied to the terminals of the apparatus in order to inject into the apparatus a current pulse having suitable characteristics, More specifically, the calibrator is designed to generate the current pulse that would be generated by the transfer of a predetermined load quantity across the terminals. Therefore, an apparent load value equal to the predetermined load value is assigned to a signal detected by the apparatus corresponding to the current pulse injected by the calibrator. The ratio between the predetermined apparent load value and the value of the amplitude of the detected signal constitutes the required calibration factor.
The discharge signals detected on an electrical apparatus are therefore associated with corresponding apparent charge values by multiplying the amplitudes of the signals by the calibration factor previously determined for that apparatus during calibration.
In light of this, it should be noted that the amplitude of the signals detected and the assignment to these signals of corresponding apparent charge values is influenced by the detection bandwidth.
It should therefore be noted that the signals detected by instruments having a very wide bandwidth cannot be compared in amplitude with signals detected on the same apparatus by instruments having a narrow bandwidth, even if each instrument has been calibrated.
Indeed, standards on the subject of partial discharge detection specify the passband values admissible for the input stages of measuring instruments.
For example, at European level, IEC 60270 lays down specifications for detection instruments in order to make the measurement results comparable and consistent. This standard thus specifies the characteristics of PD measuring instruments and also specifies how to test their performance.
More specifically, that standard lays down the specifications for the passband of the detection instruments; the standard recommends that low cut-off frequency (f1) should be between 30 and 100 kHz, that high cut-off frequency (f2) should be less than 500 kHz and that the bandwidth value (f2-f1) should be within the range of between 100 and 400 kHz, these frequency values to be measured at an attenuation of 6 dB relative to the band centre.
Accordingly, the standard states that instruments with a particularly wide detection bandwidth (greater than a few MHz) cannot be calibrated in compliance with the standard.
Thus, on the one hand, there is the need for a detection instrument having a detection bandwidth that is as wide as possible, so as to allow subsequent diagnostic processing (for example to distinguish between discharge pulse waveforms from disturbance signals or to separate signals relating to discharges that occur at different discharge sites); and on the other hand, there is a need to use an instrument with a limited detection bandwidth in order to quantify the signals detected in a way that complies with the standard.
In light of this, prior art solutions (known for example from patent documents U.S. Pat. No. 6,313,640 and EP0813282) teach the use of either broadband or narrowband instruments. More specifically, analogue filters are used which can be applied to the input stage of a broadband instrument to adapt it to the standard, thereby converting it, in practice, into a narrowband instrument.
In other cases, two instruments in parallel or in series are used, one broadband and one narrowband, but without the possibility of comparing the results in compliance with the standard and with evident waste of time and resources.
The problem of simultaneous measurement of pulses in broadband (to maintain the information useful for diagnostic processing) and in narrowband (to be able to assign to the signals a pC amplitude value that complies with the standard, that is to say, that can be compared with results of other instruments) is also dealt with in patent document WO2009/013640 to the same Applicant. In that document, a digital passband filter implemented via software is used to convert the digital discharge signal detected in broadband into a digital signal corresponding to the digital signal that would have been detected if the input stage had the desired bandwidth (compliant with the standard). This filtered digital signal is sent to an output stage together with the unfiltered digital signal so that the output stage can compare the data of the two signals (filtered and unfiltered) relating to the same pulses.
The latter solution, too, is not free of disadvantages, however.
In effect, a digital filter of that kind is complex to make and also requires expensive components.
Another problem in measuring partial discharges lies in the fact that the signals detected do not always correspond to the partial discharges but often include components corresponding to noise.
In particular, the measuring circuit often includes signal components relating to common mode noise (consisting of homopolar components) due to noise which couples to the measuring circuit through earth.
In light of this, attention is drawn to the following with regard to the partial discharge measuring circuit.
The measuring circuit includes a capacitor (called coupling capacitor) connected in parallel to the measuring impedance device to form a low-impedance grid for the signals with a high frequency content.
The measuring circuit also comprises a measuring impedance device across whose terminals the measurement signal is picked up.
The measuring impedance device is usually connected in series with the electric apparatus (in which case it is called direct-measuring impedance device), that is to say, it is connected across a low-voltage terminal of the electric apparatus and an earth node. Alternatively, the measuring impedance device is connected in series with the coupling capacitor (in which case it is called indirect-measuring impedance device), that is to say, it is connected across the earth node and a low-voltage terminal of the coupling capacitor).
Several methods are known which are used to try to “clean up” the electrical discharge signal by removing the noise components from it to leave only the components relating to the partial discharge pulses.
Some methods (described by the Applicant for example in patent document WO2007/144789) involve complex calculations to be applied to the acquired data.
Thus, for each pulse detected, it is necessary to extract one or more shape parameters and, as a function of these, to separate the acquired signals to form groups of signals that are uniform in terms of signal waveform. This is based on the assumption that the waveform of a detected signal is correlated with the source that generated the signal itself (through the transfer function that signal is subjected to as it is transferred from the source to the detection site).
These methods may be quite effective but have the disadvantage of requiring considerable resources in terms of computation capacity and time. Moreover, these methods necessitate acquiring the signals with an ultra wide band (more than ten MHz, for example) to be able to extract these shape parameters in a significant manner.
Also known (for example from DE3635611A1) are analogue partial discharge measuring instruments which allow a “transitional” measurement to be made, that is to say a differential measurement between the signals picked up at the terminals of the direct- and indirect-measuring impedance devices. These instruments make it possible to eliminate certain types of disturbances but have the disadvantage of not allowing detected signals to be processed for the purposes of extracting parameters other than the pulse amplitude and the pulse phase relative to the supply voltage. Also, this instrument necessarily has a very narrow bandwidth (a few hundred kHz) in the input stage. That means there is the risk of discharge pulses with a very high frequency content not being detected at all by the system.
Thus, the instrument is not very reliable for diagnostic purposes.